I think it's a bit suck it and see. If the software does it's sums correctly it should be damped. The swing in current needed around the crossove f - pass. The parts are cheap and part aim is to try a small |DML panel rather than a tweeter. The crossover F can be a lot lower then. Aim PC speakers, sort of mini tower 10L box. 0.353 ft^3
There is one other plot that shows what is going on at the drivers
Boxsim can also handle active crossovers. I think that is the next step. I should have done this sort of thing before I bought the bass drivers. That would probably push me up to a 5". Also I could then knock up a case according to the paper design. Fit the speakers, measure the parameters and the load them via as per the box design rather than infinite/din baffle.
The difficulty in using Boxsim in my view is the optimiser. It has sliders to influence it's behaviour but also tries to optimise all including phase. With a mix of cone speakers crossovers may finish up anywhere and may look a bit odd but phase ok. A tweeter has such a sharp cut off the crossover f is pretty predictable.
Best add I posted this as an example of the use of Boxsim.
A problem I didn't mention is max spl. Use Visaton's suggested +/-2mm cone excursion when the driver is loaded and max spl dips to 80db at 45Hz. It goes up to 90 odd at 125Hz. It's higher at 30Hz - box tuning.
They don't state a suitable figure on the data sheet. They do mention 7mm as the max cone movement on the web. Seems that is +/- going on the actual drivers.
Increase the +/-2 and more power but what power levels are needed for near field on a PC? Effects on distortion? LOL Looks like it's worth building one to find out.
A problem I didn't mention is max spl. Use Visaton's suggested +/-2mm cone excursion when the driver is loaded and max spl dips to 80db at 45Hz. It goes up to 90 odd at 125Hz. It's higher at 30Hz - box tuning.
They don't state a suitable figure on the data sheet. They do mention 7mm as the max cone movement on the web. Seems that is +/- going on the actual drivers.
Increase the +/-2 and more power but what power levels are needed for near field on a PC? Effects on distortion? LOL Looks like it's worth building one to find out.
Silly me again >
I was thinking the woofer was 8 ohms. It all gets very embarrassing 😕
If I only had a dollar for every time I've done that 😉
However the key to an effective circuit is the shape vs the topology.. and one thing optimisers usually don't do is to add to the topology. It's back to us to say hey, the optimiser's stuck again, what can I add to this.
I'm yet to find an optimiser that can do a better job than a person.. although rarely I'll use one to find a starting point for something unusual.
However the key to an effective circuit is the shape vs the topology.. and one thing optimisers usually don't do is to add to the topology. It's back to us to say hey, the optimiser's stuck again, what can I add to this.
Yes the tricky part. I've had a long term interest in speakers back to the time when people would use a calculator. Design an 20Hz F3 horn that wouldn't fit in the house etc, Build based on this no way, no confidence and how to measure performance. Things are easier now so the interest resurfaces. Crossovers - maths no fun, that's what software is for. 😉 I'm finding the optimiser educational in this area.
Looking around on the web I found 2 ideas on developing crossovers. One I suspect is rather traditional. Compensate the driver impedances and add zobels if needed - then the crossovers. Another using software I wouldn't buy. It shows the idealised slope and what the crossover does - add bits to get the correct slope. That needs experience but an optimiser can show what the bits do.. The video covers a high pass. Mentions doing a video for woofers but hasn't so I have to wonder what can be added. 😉 It's probably trickier.
Where I have fell out with it a bit is using a cone speaker for HF. Small ones do not beam as much. Also tried a concentric. There is no control of crossover F. It will seek perfection which doesn't mean that the required F wouldn't be acceptable. There may be ways around this but say one speaker has a higher spl. Unless that isn't corrected first a partial optimisation will flatten things with an eventual slope up to that spl.
There is also a sort of forum problem. Someone who seems to be an expert reckons that for near field the cross over needs to be as high as possible. They would use rather expensive speakers. I might at some point but on the other hand how good do they need to be for me. My aim anyway is no need for a subwoofer.
2 special "speakers" can be added to Boxsim so they can be loaded. A voltmeter and an ammeter.
😉Maybe I should look at a 3 way.
I'm not familiar with what constraints Boxsim requires, ideally a full plot definition would be good, but in reality there's a lot you can do specifying level, crossover frequency, slope and Q, plus perhaps tilt.
I believe it's worthwhile to define your own frequencies based on your design acoustic conditions.. a crossover optimiser doesn't know about those, and has no idea of acoustics unless perhaps it uses polar measurements, and even then I'd make as many decisions for it as I could.
This is a little on filter shape vs circuit. I don't know if it would be helpful to you but I thought I'd offer it - https://www.diyaudio.com/community/...overs-without-measurement.189847/post-6910468
Thanks. A bit of help. The video suggested use of the zobel like networks after a fashion. I'll scan through all of the posts in the thread.
I watched an Xsim demo video. Boxsim's calcs are too slow to allow that sort of set up. At the end of the video he looked at power. 100w in - 10w out which to me means do the entire thing again. It seems that crossover design is likely to be tedious how ever it's done.
Zobel type compensation can be a helpful thing. I sometimes use it when I know I'll have to fix some aspect to make another variable, or perhaps more versatile. A good example is taking out a dome tweeter resonance peak. It serves no useful purpose and with it out of the way, it makes it easier to focus on filtering.
RC on a woofer is probably not as necessary if you know you plan to use a second order filter, since those act like a self contained compensating network at the extremes of their band. You can train yourself to see this on a simulator, make comparisons with ideal versions of drivers vs real impedances.
You can also see which small adjustments give you the same response with and without the Zobel, since it is possible to do the same thing with and without impedance compensation in most cases.
RC on a woofer is probably not as necessary if you know you plan to use a second order filter, since those act like a self contained compensating network at the extremes of their band. You can train yourself to see this on a simulator, make comparisons with ideal versions of drivers vs real impedances.
You can also see which small adjustments give you the same response with and without the Zobel, since it is possible to do the same thing with and without impedance compensation in most cases.
I finally found out how to connect the voltmeter - I thought it would be easy. Just connect it to the points that need measuring. Hopefully it will indicate the effect of impedance peaks. Disconnect if optimising. No info on the ammeter. The only place I could find any details was the German Visaton forum. Pity these can't be translated.
L pads. I've noticed use of rather low resistor values. There is a calculator that can specify the resultant resistance based on the resistance of the driver and dB loss needed. It gives the R's needed. A series resistor before the cross over is another option. Maybe even after it.
Optimiser sliders. The one that favours min impedance levels is pretty obvious . Another appears to relate to peaks. The power one gave a 5db increase on the woofer at max cone excursion by increasing it's preference level.
The acoustic power slider. I don't know anything about this metric. It's plotted and generally falls over the frequency range. All seem to concentrate on SPL so is this plot important? Music is usually mixed to a pink profile or close to it.
So tried a 3 way. I may have found a bug. The project settings can be edited to say add a speaker. I added the mid having started with 2 speakers. L pads ok but impedance correction refused to do anything. Or maybe it's a feature of the package or the speaker. It appears to run under Linux via wine without any problems. There is a need to shut windows that are up correctly using ok other than those provided under extras.
No help in the package for mid crossovers. What I have done with say correcting a HF response is looked at designs. Added bits and optimised them 😉 shows me what they do. 🙂 I'll try the same with a mid crossover.
The optimiser can save the current state as a reference. This is plotted dotted and the changes are shown solid as it works away. The results can be applied updating the schematic so those changes can be studied, The reference can be restored if needed. It may optimise parts out via showing ridiculously low numbers. Remove them and optimise again. LOL with a silly crossover for the mid it optimised it out completely and finally bought it back - 🙂 at very very low levels.
L pads. I've noticed use of rather low resistor values. There is a calculator that can specify the resultant resistance based on the resistance of the driver and dB loss needed. It gives the R's needed. A series resistor before the cross over is another option. Maybe even after it.
Optimiser sliders. The one that favours min impedance levels is pretty obvious . Another appears to relate to peaks. The power one gave a 5db increase on the woofer at max cone excursion by increasing it's preference level.
The acoustic power slider. I don't know anything about this metric. It's plotted and generally falls over the frequency range. All seem to concentrate on SPL so is this plot important? Music is usually mixed to a pink profile or close to it.
So tried a 3 way. I may have found a bug. The project settings can be edited to say add a speaker. I added the mid having started with 2 speakers. L pads ok but impedance correction refused to do anything. Or maybe it's a feature of the package or the speaker. It appears to run under Linux via wine without any problems. There is a need to shut windows that are up correctly using ok other than those provided under extras.
No help in the package for mid crossovers. What I have done with say correcting a HF response is looked at designs. Added bits and optimised them 😉 shows me what they do. 🙂 I'll try the same with a mid crossover.
The optimiser can save the current state as a reference. This is plotted dotted and the changes are shown solid as it works away. The results can be applied updating the schematic so those changes can be studied, The reference can be restored if needed. It may optimise parts out via showing ridiculously low numbers. Remove them and optimise again. LOL with a silly crossover for the mid it optimised it out completely and finally bought it back - 🙂 at very very low levels.
An ammeter is normally connected in series. You break the circuit and insert it. Be careful not to pass too much through it.
Power does fall with frequency when the response is flat.
It's interesting that impedance compensation doesn't work. Could you show the circuit in question?
Power does fall with frequency when the response is flat.
It's interesting that impedance compensation doesn't work. Could you show the circuit in question?
Nothing else connected other than the amp. A Visaton W100S-8 which may be a bad choice. I need to go through it again to make sure but the impedance curve didn't change at all when I added it. It may be due to something I did as well as the other factors I mentioned. I also did the schematic a different way. The application creates one per amp used so the amp symbol can be shown on each speaker network rather than linking all to the same symbol. First time I have done that.
In a sense, technically when you connect it directly across the amp nothing does change.
However simulators usually show the impedance looking out from the amp.
However simulators usually show the impedance looking out from the amp.
I simply connected the speaker to the amp and added a RC across the speaker expecting the speakers impedance curve to change as it has with others.
The package will also calculate a couple of other compensations. RLC and a more complex version of that, Still no change.
😉 I'm hoping this is just down to starting a project with 2 speakers and when those were sorted editing the project settings to 3. Or some sort of silly by me. As I can run 2 or more instances it's pretty easy to start again with a 3 speaker project and manually copy stuff across but I will try adding it again on the current one. I need to revise the crossover settings anyway so will remove those parts and try connecting the mid to one of the other amp symbols rather than adding another.
The package will also calculate a couple of other compensations. RLC and a more complex version of that, Still no change.
😉 I'm hoping this is just down to starting a project with 2 speakers and when those were sorted editing the project settings to 3. Or some sort of silly by me. As I can run 2 or more instances it's pretty easy to start again with a 3 speaker project and manually copy stuff across but I will try adding it again on the current one. I need to revise the crossover settings anyway so will remove those parts and try connecting the mid to one of the other amp symbols rather than adding another.
That was the problem. I was looking at the selectable speaker plot which just shows the actual speaker impedance. To look at what the amp sees I need to use the main plot and disconnect the other speakers from the amp. That plot shows the effects of the baffle. The selectable speaker one also shows a plot in a 1/2 room indicating bass boost also the baffle effect.
This is what the optimiser did to an RLC compensation. The calc'd one is shown dotted,
Yes 😉 Entering those values driver in a 2L sealed box gives this
And the optimiser wrecks it what ever selection of slider settings I set. The problem seems to be it trying to do something about the box and can't so goes completely AWOL. I can set it as a tweeter - no box which removes the kink but it still goes AWOL. Oddly if I set valve amp and 4ohm transformer it improves the boxless curve a bit. Also with the box. It can also be set up to favour a set frequency range. Looks like it wont totally ignore the rest.
Not sure what to make of selecting valve amp and a transformer impedance. It looks like it may optimes the impedance as flat as it can be then - 🙁 subject to the number of bits added. The tricky aspect.